0 Blackwell Publishers Ltd. 1996. 108 Cowley Road, Oxford, O X 4 l J F , UK
and 238 Maiti Street, Cambridge, M A 02142, USA.
Mind G. Language, ISSN: 0268-1064
Vol. 11. No. 2 June 1996, pp160-190.
Evidence for the Innateness of Deontic Reasoning
DENISE DELLAROSA CUMMINS
Abstract: When reasoning about deontic rules (what one may, should, or should
not do in a given set of circumstances),reasoners adopt a violation-detection strategy,
a strategy they do not adopt when reasoning about indicative rules (descriptions of
purported state of affairs). I argue that this indicative-deonticdistinction constitutes
a primitive in the cognitive architecture. To support this claim, I show that this distinction emerges early in development, is observed regardless of the cultural background
of the reasoner, and can be selectively disrupted at the neurological level. I also argue
that this distinction emerged as a result of selective pressure favouring the evolution
of reasoning strategies that determine survival within dominance hierarchies.
If seven decades of empirical investigations of human reasoning have shown
us nothing else, they have shown us that our reasoning strategies vary as a
function of problem content (Byrne, 1989; Cheng and Holyoak, 1985, 1989;
Cheng, Holyoak, Nisbett and Oliver, 1986; Cosmides, 1989; Cummins, 1995,
in press; Cummins, Lubart, Alksnis and Rist, 1991; Evans, 1989; Griggs and
Cox, 1982, 1983; Henle, 1962; Manktelow and Over, 1991; Revlin and Leirer,
1978; Roberge, 1982; Thompson, 1994, 1995; Thompson and Mann, 1995;
van Duyne, 1974; Wason, 1968; Wason and Johnson-Laird, 1972; Wason and
Shapiro, 1972; Wilkins, 1928). While robust, pervasive, and readily evoked,
the significance of content effects is far from clear. Initial explanations attributed them to the idiosyncratic influence of an individual’s knowledge on
a content-free, syntactically-driven reasoning process (Braine, 1978; Braine
and OBrien, 1991; Henle, 1962; Rumain, Connell and Braine, 1983). In recent
years, however, a groundswell of opinion has arisen that it is in the evocation
of content effects that the true nature of the human reasoning architecture
is to be found. Human reasoning varies as a function of content because
the human reasoning architecture consists of a collection of domain-specific
Portions of this paper were presented at the Conference for Epistemiology and Evolutionary Psychology, Rutgers Unversity, April, 1995. I thank Nick Chater and David Over for
their invaluable comments on an earlier draft of this paper.
Address for correspondence:Cognitive Science, Psyihdogy 312, University of Arizona,
Tucson, AZ 85721, USA.
Email: [email protected].
Deontic Reasoning Modules
161
acquired schemas (Cheng and Holyoak, 1985, 1989; Cheng et al., 1986),
evolved modules (Cosmides, 1989; Cosmides and Tooby, 1989, 1992, 1994),
or model-building strategies (Johnson-Laird and Byrne, 1991; Manktelow and
Over, 1991) rather than a collection of syntactically-driven rules. For this reason,
different contents evoke different domain-relevant reasoning strateges.
1. The Indicative-Deontic Distinction in Adult Reasoning
One of the most robust content effects in the literature is what I will call
the indicative-deontic distinction. Indicative reasoning is reasoning about the
epistemic status of rules. It can be analyzed in terms of hypothesis-testing,
that is, testing the truth content of a rule that describes a purported state of
affairs. Deontic reasoning, on the other hand, is reasoning about what one
may, ought, or must not do in a given set of circumstances (Hilpinen, 1971,
1981; Manktelow and Over, 1991). Virtually all of our social institutions presuppose a capacity to understand and reason about what is permitted, obligated, prohibited, cautioned, or advised. Failure to reason effectively about
deontic rules can have disastrous consequences, including scolding, expulsion, legal action, and even incarceration. We also presume the capacity to
reason deontically in most of our child-rearing and social interactions,
appealing to this presumed capacity whenever we utter a permission, promise, warning, or threat.
Human reasoners treat these two types of rules very differently, adopting
a violation-detecting strategy when reasoning about deontic rules and a confirmation-seeking strategy when reasoning about indicative rules. The clearest examples of the indicative-deontic distinction are based on the Wason
card selection task, a task unique not only in its simplicity but in its ability
to generate robust content effects (Wason, 1968). This task consists of asking
reasoners which of four cards must be turned over in order to test a particular conditional rule (p 4). The four cards correspond, respectively, to
the antecedent of the conditional (p), its consequent (q), and the denial of
each (not-p, not-q). For example, consider the following problem: A friend
relates to you the observation that in Arizona ’If you go to Phoenix, you
travel by train.’ In front of you are four cards that have a person’s destination
on one side and his or her means of transportation on the other. Your task
is to indicate all and only those cards that must be turned over in order to
test whether or not your friend was telling the truth. Your choices are ’Phoenix’, ‘Tucson’, ’Train’ and ’Car’. If you are like the vast majority of people,
you selected ‘Phoenix‘ and ‘Train‘, that is, p and q.
Now consider the following case. You are to pretend that you work for
the Arizona transportation bureau, and it is your job to enforce a new law
aimed at reducing air pollution due to car emissions. The law is ’If you go
to Phoenix, you must travel by train.’ You’re shown the same four cards,
and are asked to indicate all and only those cards that must be turned over
in order to determine whether or not the rule is being followed. If you’re
-
0 Blackwell Publishers Ltd. 1996
Mind 6 Language
162
% pand
not-4
choices
(a)
"Abstract" Content Conditionals
6% If there is a 3 on one side of any card, then there is a D on its other side.
(Wason, 1968, Exp. 1)
8% If there is a square on one side, then there is a red scribble on the other
side. (Wason, 1968, Exp. 2)
29% Every card which has a vowel on one side has an even number on the
other side. (Wason and Shapiro, 1971, Exp. 1)
12% Every card which has a D on one side has a 3 on the other side. (Wason
and Shapiro, 1971, Exp. 2)
4 70 If there is a vowel on one side of the card, then there is an even number
on the other side of the card. (Wason and Johnson-Laird 1972)
8% If a letter has A on one side, then it has a 3 on the other side. (JohnsonLaird, Legrenzi and Legrenzi, 1972)
8% Every card which has a D on one side has a 5 on the other side.
(van Duyne, 1974)
12% If a card has a D on one side then it has a 5 on the other side. (van Duyne,
1974)
0 4 % If a card has an ' A on one side, then it has a 5 on the other side. (Griggs
and Cox, 1982, Exp. 1-3)
4 70 If a card has an A on one side, then it has a 3 on the other side. (Cox and
Griggs, 1982, Exp. 1)
5% If a card has an 'A' on one side, then it has a 5 on the other side. (Griggs
and Cox, 1983, Exp. 1 )
0% If a card has a vowel on its letter side, then it has an even number on its
number side. (Platt and Griggs, 1983)
19% If there is an ' A on one side of the card, then there is a "4" on the other
side of the card. (Cheng and Holyoak, 1985, Exp. 2)
Figure 2 Some examples of conditional rules used in studies of adult human
reasoning, and average reported performance. The conditionals in (a) are
abstract in that fhey do not describe situatiorzs with which most people are
familiar. Average performance is uniformly low. The conditionals in (b) are
neutral in content; they embed familiar situations, but they do not convey
social regulations. Average performance does not differ appreciably from that
observed on abstract conditionals. The conditionals in (c) v a y in terms of
familiarity and abstractness, yet they all describe permission situations. Average performance on these social regulations is uniformly high.
like most people, it seems apparent now that 'Phoenix' and 'Car', that is, p
and not-q must be turned over.
The travel problem is an example of an indicative rule, and the most frequently observed response pattern (p and q) constitutes seeking rule-confirming evidence. The law problem is an example of a deontic rule, and the
most frequently observed response pattern (p and not-q) constitutes seeking
rule violations. This is what I mean by the indicative-deontic distinction in
human reasoning. Its robustness is readily apparent in Figure 1(a)-(c).
Figure l(a) shows the low level of violation checking observed when
0 BIackwell Publishers Ltd. 1996
Deonfic Reasoning Modules
163
% pand
(b)
Neutral Themafic Content Conditionals
not-q
choices
62% Everytime I go to Manchester, I travel by car. (Wason and Shapiro, 1971,
Exp. 2)
8% Everytime Ottawa is on one side, car is on the other side (Bracewell and
Hidi, 1974)
22% Everytime I go to Manchester, I travel by car. (Gilhooly and Falconer,
1974)
9% Everytime I go to Miami, I travel by car. (Griggs and Cox, 1982)
25% If a person is wearing blue, then the person must be over 19. (Cox and
Griggs, 1982, Exp. 1)
54% If a person is under 19, then the person must be drinking Coke. (Cox and
Griggs, 1982, Exp. 2)
20% If a person is over 19, then the person must be drinking beer. (Cox and
Griggs, 1982, Exp. 3)
4% If I eat haddock, then I drink gin. (8 food and drink combinations tested)
(Reich and Ruth, 1982, Exp. 1)
12% When I go to work, I hurry/When I travel to France, I go by planefwhen
the fruit are yellow, they are ripe/When it is early, Molly serves tea.
(Reich and Ruth, 1982, Exp. 2)
17% If a bird on the island has a purple spot under its wing, then it makes its
nest on the ground, (Cheng and Holyoak, 1985, Exp. 3)
60% If an envelope is sealed, then it must have a 20 cent stamp. (Cheng and
Holyoak, 1985, Exp. 1) (When presented without accompanying permission
rationale.)
60% If a passenger form says ’Entering’ on one side, then the other side must
include ‘cholera’ (Cheng and Holyoak, 1985, Exp. 1) (When presented
without accompanying permission rationale.)
Figure 1 (b).
unfamiliar, neutral contents are embedded in the task, from 0% to about
25%. Figure l(b) shows an equally low incidence level when familiar neutral
contents are embedded. But a dramatically different picture appears in Figure l(c); here violation detection selections range from 50% to 96%, substantially higher than on the equally familiar but neutral content problems in
Figure l(b). What the problems in Figure l(c) have in common is this: they all
embed permission and obligation contents, and hence, by definition, require
deontic reasoning. The indicative-deontic distinction is also readily evoked
among reasoners of varying educational backgrounds (Cheng, Holyoak, Nisbett and Oliver, 1986), and on other reasoning tasks, such as conditional
arguments (Fillenbaum, 1978; Thompson, 19941, paraphrasing (Fillenbaum,
1975, 1976; Thompson and Mann, 1995) and equivalence judgments
(Fillenbaum, 1976).
Numerous proposals have been put forth to explain the indicative-deontic
distinction. The first is pragmatic reasoning schema theory (Cheng and
Holyoak, 1985, 1989; Cheng, Holyoak, Nisbett and Oliver, 1986). According
to this theory, adults excel at deontic reasoning because deontic concepts
0 Blackueff Publishers Ltd. 1996
164
Mind & Language
% pand
(C)
not-q
Deontic Content Rules
choices
88% If a letter is sealed, then it has a 50 lire stamp on it. (Johnson-Laird,
Legrenzi and Legrenzi, 1972)
75-96% If a person is drinking beer, then the person must be over 19. (Cox and
Griggs, 1982, Exps 1-3)
72% If a person is drinking beer, then the person must be over 19. (Griggs and
Cox, 1982, Exp. 3)
70% If a person is drinking beer, then the person must be over 19. (Griggs and
Cox, 1983, Exp. 1)
50% If a student studies philosophy, then he is at Cambridge. Embedded in
scenario involving eligibility for grant support (van Duyne, 1974)
58% Every student who studies physics is at Oxford. Embedded in scenario
involving eligibility for grant support (van Duyne, 1974)
85% If a purchase exceeds $30, then the receipt must have the signature of the
departmentstore manager on the back. (Griggs and Cox, 1983, Exp. 1)
90% If an envelope is sealed, then it must have a 20 cent stamp. (Cheng and
Holyoak, 1985, Exp. 1)
(When presented with accompanying permission rationale.)
90% If a passenger’s form says ’Entering’ on one side, then the other side must
include ‘cholera’ (Cheng and Holyoak, 1985, Exp. 1) (When presented with
accompanying permission rationale.)
61% If one is to take action A, then one must first satisfy precondition P (Cheng
and Holyoak, 1985, Exp. 2)
75% If a man eats cassava root, then he must have tattoo on his face.
(Cosmides, 1989, Exp.
89% If someone stays overnight in the cabin, then that person must bring along
a bundle of wood from the valley. (Gigerenzer and Hug, 1992, Exp. 2)
63% If you tidy your room, then I will let you go out to play. (Manktelow and
Over, 1991, Exp. 2)
42 % If the purchases exceed 10,000 francs, then the salesman must stick on the
back of the receipt a voucher gift for a gold bracelet. (Plitzer and NguyenXuan, 1992, Exp. 1)
Figure 1 (c).
constitute classes of frequently-encountered situations for which collections
of domain-specific, goal-orientated rules are induced. One such schema, the
permission schema, details the relationship between actions and preconditions, such as ’If the precondition is satisfied, then the action may be taken.’
A second theoretical explanation is social exchange theory, which analyzes
deontic reasoning in terms of cost/benefit analysis and cheater detection
(Cosmides, 1989; Cosmides and Tooby, 1994).These strategies are proposed
to be innate, having been selected for during the evolution of our species in
order to reason effectively about social exchange (cooperativeaction for mutual benefit). A third theory explains the deontic effect in terms of the construction and manipulation of models based on subjective utility (Manktelow
and Over, 1991, 1995). Finally, a fourth theory models performance on the
selection task in terms of optimal data selection using decision theory
0 B l a c h e l l Publishers Ltd. 1996
Deontic Reasoning Modules
165
(Oaksford and Chater, 1994). In the indicative case, reasoners choose to
inspect instances that are expected to yield the most information about which
of two competing hypotheses are true (i.e. ‘if p is true, then q must be true
as well’ or ’p and q are independent’). According to this model, the expected
information gain for the q card is greater than the expected information gain
for the not-q card. In the deontic case, on the other hand, reasoners choose
instances with an eye toward maximizing expected utility. In this case, a
violation-detection strategy maximizes the expected utility function. The
indicative-deontic distinction therefore indicates that human reasoning is
optimally adapted to the environment, or domain, to which it is applied.
Despite the very considerable differences among these theoretical explanations, they all have two things in common. The first is that human reasoning strategies are domain specific; people do not - and ought not to - use
the same strategies when reasoning about indicative and deontic contents.
The second commonality is the central role afforded violation detection in deontic
reasoning. There is general agreement among descriptive and normative theorists that a crucial part of reasoning deontically is appreciating the necessity
of detecting violations of deontic rules. For example, in the case of permissions, one must ensure that no one has taken a specified action unless
specified preconditions have been satisfied (e.g. ‘If you want to take a book
out of the library <permitted action>, you must have a valid library card
<condition>’.) In the case of obligations, one must ensure that no one has
avoided doing what is obligated under the specified circumstances ( e g ’If
you lost a library book <condition>, you must pay $25 in fines <obligatory
action>.’) In the case of prohibitions, one must ensure that no one has done
something forbidden (e.g. ’No one may slide down the slide backwards.’)
In contrast, what constitutes optimal performance on indicative reasoning
tasks (where reasoners are asked to test the truth of a rule) is the subject of
some controversy. Traditionally, violation detection played a central role in
normative theories of indicative reasoning because observing a violation of
an indicative rule (hypothesis) disproves the rule (e.g. finding a white raven
disproves the rule ‘all ravens are black‘) and hence provides incontrovertible
proof of the rule’s truth content (Popper, 1959). Seeking confirming evidence,
the typical strategy employed by adults on indicative reasoning tasks, was
considered at best a bias in the reasoning process and at worst an error
because a confirming instance does not provide incontrovertible evidence
about the rule’s truth content (Evans, 1989; Wason, 1968).Many contemporary philosophers of science reject this exclusive emphasis on falsification
strategies (e.g. Churchland, 1986; Federov, 1972; MacKay, 1992; Putnam,
1974; Quine, 1953), as does the decision-making analysis of card selection
performance described above (Oaksford and Chater, 1994).
However one measures the normative value of these strategies, the fact
remains that when reasoning about deontic rules, adults spontaneously
adopt a violation-detection strategy, and when reasoning about indicative
rules, they spontaneously adopt confirmation-seeking strategies. There is
agreement among theorists that this distinction is a direct reflection of innate
0 Blackwell Publishers Ltd. 1996
166
Mind 6 Language
or acquired domain-specificity in the human reasoning process. I will take
this claim one step further and argue for the existence of a domain-specific
module devoted exclusively to deontic contents.
At the heart of my position lies an evolutionary argument: evolutionary
theory is based on the assumption that there is a causal relationship between
the adaptive problems a species repeatedly encounters during its evolution
and the design of its phenotypic structures. The structure in question here
is a functional one-the human reasoning architecture. My position is that
the most pressing adaptive problems primates (and probably all other social
creatures) faced during their evolution were ones of within-species social
coordination and social interaction, producing enormous pressure to
develop strategies to solve these problems. The colonial insects solved them
by evolving members of different social ‘castes’, such as queen bees and
worker bees. Mammals seem to have solved them by evolving cognitive
architectures that enable them to be good social reasoners. In contrast, there
was no corresponding pressure to be a good scientific reasoner. In fact, it
was not until the evolution of language and the symbolic representation
schemes it afforded (e.g. mathematics) that scientific reasoning emerged.
From this perspective, social reasoning was primary, and scientific reasoning
is a late-emerging capacity.
More specifically, I will argue that a domain-specific deontic reasoning
module evolved for the very important purpose of solving problems that
frequently arise within a dominance hierarchy-the social structure that
characterizes most mammalian and avian species. Remaining within the
social group reduces the risk of death due to predation, increases the chances
of reproductive success (due to greater access to potential mates and mating
opportunities), a s well as increasing the viability of survival of the young.
Using data from primatological studies, I will argue that remaining and surviving within a dominance hierarchy depends crucially on the capacity to
detect and respond appropriately to permissions, obligations, prohibitions,
promises, threats and warnings. Failure to do so carries a high risk of inciting
agonistic encounters or ostracism. The core component of this domain-specific module is violation-detection: to reason effectively about deontic concepts, it is necessary to recognize what constitutes a violation, respond to it
appropriately (which often depends on the respective status of the parties
involved), and appreciate the necessity of adopting a violation-detection
strategy whenever a deontic situation is encountered.
I will argue that this analysis explains a variety of data, including (a) why
people access different strategies when reasoning about deontic and indicative situations, (b) why violation detection is the preferred strategy in deontic
tasks, (c) why that preference emerges early in development, and (d) why
it is observed regardless of very substantial cultural and educational differences among reasoners. I will also present evidence from neuropsychology
suggesting that social reasoning and decision-making is dissociable at the
neuropsychological level from other types of intelligent reasoning, thereby
producing the ’smoking gun’ of phenotypic structures that evolved in
0 Bhckwell Publishers Ltd. 1996
Deontic Reasoning Modules
167
response to these hypothesized evolutionary pressures. Finally, I will consider in greater detail alternative explanations for the indicative-deontic distinction including rule-based theories (Braine and OBrien, 1991; Osherson,
1974,1975; Rips, 19941, pragmatic schema theory (Cheng and Holyoak, 1985,
19891, social exchange theory (Cosmides, 1989; Cosmides and Tooby, 19941,
utility theory (Manktelow and Over, 1991, 1995), and optimal data selection
theory (Oaksford and Chater, 1994).
2. Deontic Reasoning in Non-Human Primates
Homo sapiens is a primate species whose ancestors diverged within the primate line a scant five million years ago. At a molecular (DNA) level, humans
and chimpanzees are only 1%different. Because evolution builds upon existing structures, it seems reasonable to assume that certain characteristics of
human reasoning may have roots that reach deep into our evolutionary past,
prior to the splitting of hominids within the primate line.
The most striking of these characteristics is that, like human reasoning, the
reasoning of non-human primates is subject to content effects. For example,
squirrel monkeys and chimpanzees can perform transitive inference on
object-oriented tasks only after considerable drilling with paired stimuli
(Gillan, 1981; McGonigle and Chalmers, 1977).Yet they readily make transitive inferences while making complex kin and dominance rank discriminations among individuals in their social groups (Dasser, 1985, pp. 16-19;
Cheney and Seyfarth, 1990, pp. 91-96). Within their social groups, they also
evidence an appreciation of causality and reciprocity that is not apparent in
their dealing with physical objects (Cheney, 1978; Datta, 1983a+; Seyfarth,
1981). So pervasive is this 'social content effect' that the nature of primate
intelligence is generally believed to have been shaped by the exigencies of
life within the social group (Cheney and Seyfarth, 1985, 1988, 1990;
Humphrey, 1976; Jolly, 1966; Whiten and Byrne, 1988a).As Cheney and Seyfarth (1985, p. 39) put it, 'among primates, evolution has acted with particular force in the social domain'.
So what exactly are the exigencies of life within non-human primate social
groups? Their social interactions are characterized by (a) dominance
relations and (b) coalition and alliance formation. In fact, non-human primates have been described as consummate tacticians, with much of this tactical reasoning aimed at jockeying for position within the dominance hierarchy (e.g. Whiten and Byme, 1988a, 1988b; Harcourt and de Waal, 1992).
In functional terms, a dominance hierarchy is simply the statistical observation that 'particular individuals in social groups have regular priority of
access to resources . . . in competitive situations' (Clutton-Brock and Harvey,
1976, p. 215). In its most developed form, it is transitive, meaning that if A
has priority over B, and B has priority over C, then A has priority over C,
and so on. The role of dominance is most pronounced in situations characterized by high levels of competition for resources, such as high population
0 Blochell Publishers Lfd. 1996
168
Mind b Language
density or the onset of breeding season (Clutton-Brock and Harvey, 1976).
Contrary to folk wisdom, dominance ranking is not correlated with size.
Instead, one's rank in the hierarchy depends crucially on the ability to form
and maintain strong alliances, the sine qua non of social skills (Harcourt, 1988;
Harcourt and DeWaal, 1992; Packer, 1977; Seyfarth and Cheney, 1984; Smuts,
1985).There is therefore a direct relation between social reasoning skill and
the ability to dominate resources and hence increase chances of survival.
The social reasoning that is required to secure and maintain a high-ranking
place within the dominance hierarchy is shot through with deontic concepts.
Consider first the concepts of permission and prohibition. Those who currently dominate resources determine who may engage in which activities
when, and they punish transgressors. For example, dominant males monopolize reproduction opportunities by aggressing against females and subordinate males who are caught socializing or consorting. De Waal (1982)
describes a dominant male whose peculiar means of punishing errant
females involved jumping up and down on them. Because of the high risks
involved in such forbidden liaisons, females and subordinate males often
engage in deception, such as concealing their trysts and suppressing their
copulation cries; subordinate males also hide their erections behind their
hands when their courtships are interrupted by dominant males (Kummer,
1988; de Waal, 1988). Deceptions of this kind have also been observed for
hiding other forbidden behaviours, such as stealing food, failing to share
food, or grooming forbidden individuals (for numerous examples, see Byrne
and Whiten, 1988; Whiten and Byrne, 1988b).For example, one baboon spent
twenty minutes inching behind a rock so that a dominant male could not
see her grooming a subordinate male.
In order to avoid agonistic encounters, it is therefore crucial to reason
effectively about what is permitted and what is forbidden. This requires, at the
very least, the capacity to classify instances into these two categories (e.g.
'won't elicit aggression' and 'will elicit aggression', respectively), and to
respond appropriately (e.g. 'indulge in the activity, if I so desire' and 'refrain
from engaging in the activity, regardless of my desires', respectively).
Discriminating between these classes does not necessarily require engaging in some activity first and then suffering the consequences; the individual
can instead attend to the warning or threat signals communicated by another
group member. These signals are typically uttered when another individual
is attempting to do something forbidden, and they are typically heeded.
Male vervet monkeys, for example, utter threat grunts at rivals who attempt
to copulate with estrus females (Cheney and Seyfarth, 1990, p. 227). Male
silverback gorillas sometimes utter warning or threat cough-grunts to juveniles who play too boisterously near them or who come too close to the
human observer (Cheney and Seyfarth, 1990, p. 227). These threat signals,
therefore, serve to communicate a prohibition, and avoiding an agonistic
encounter requires the capacity to recognize the prohibition and refrain from
engaging in the forbidden activity.
Life within existing dominance hierarchies, therefore, requires detecting
0 Blackwell Publishers Ltd. 1996
Deonfic Reasoning Modules
169
and responding appropriately to permissions, prohibitions, and warnings.
Lower-ranking individuals attempt to engage in forbidden activities in order
to secure a larger share of resources, and higher-ranking individuals defend
their privileged access to resources by detecting and punishing acts of cheating. This is the essence of permission and prohibition structures: those in
positions of authority (or dominance) determine who may engage in which
activities when, and threaten or punish transgressors.
It is in the interest of subordinates, on the other hand, to broaden their
access to available resources. In other words, it is in their interest to move
u p in rank. For example, among male primates, rank within the dominance
hierarchy is acquired and maintained through dyadic aggression, and
alliances determine the fate of outranked individuals, including alpha males
whose rank is usurped (Chapais, 1988,1992; Datta, 1983a-b; Goodall, 1986;
Harcourt and Stewart, 1987; Harcourt and de Waal, 1992; Riss and Goodall,
1977; Uehara, Hiraiwa-Hasegawa, Hosaka and Hamai, 1994). Alpha males
who form or already possess strong alliances with other males maintain a
relatively high, stable position within the group, while those who have no
alliances or weak alliances are ostracized, maintaining a solitary existence
outside the group (Goodall, 1986; Riss and Goodall, 1977; Uehara, HiraiwaHasegawa, Hosaka and Hamai, 1994; de Waal, 1982).
Alliances are therefore crucial to survival within primate troupes, and,
importantly, alliances are formed and maintained on the basis of reciprocal
obligations. Cheney and Seyfarth have reported that vervet monkeys are more
likely to respond to calls from non-kin during agonistic encounters if the
caller has groomed them recently; they also form the strongest alliances with
individuals who groom them most often (Cheney and Seyfarth, 1990, pp. 6769; Seyfarth, 1976; Seyfarth and Cheney, 1984). Reciprocal obligations, therefore, have the structure of a promise as in 'If you groom me, I'll support you
in a fight.' A promise constitutes a commitment on the part of the promiser
that becomes an obligation once the promisee has satisfied the conditions of
the commitment (e.g. 'You've groomed me, so now I must support you in
your fight'), and a permission from the viewpoint of the promisee to engage
in some activity (e.g. 'I may engage in this fight because you will support
me in return for my grooming.') (Politzer and Nguyen-Xuan, 1992)
This appreciation of obligation structures is also imbued with a 'machiavellian' sophistication: individuals prefer to groom and to assist individuals
of higher rank than themselves. This preference presumably is due to the
fact that support from higher-ranking individuals during agonistic encounters has greater effect than support from lower-ranking individuals. For
example baboons, macaques, and vervet monkeys form matrilineal hierarchies in which any female is dominant to all the females that are subordinate to her mother, and she is subordinate to all the females that are dominant to her mother (Chapais, 1992; Cheney and Seyfarth, 1990; Prud'Homme
and Chapais, 1993). During agonistic encounters, support is typically given
to the higher-ranking females who in turn intervene in conflicts when they
themselves are dominant to the target of the aggression. By aiding higher0 Blackwell Publishers Ltd 1996
170
Mind b Language
ranking females, lower-ranking females form strong alliances based on
reciprocal obligations that enable them to move up in rank.
If changing rank crucially depends on reciprocity, then effective reasoning
about obligations requires that violations of reciprocity yield negative consequences for the cheater. There is some evidence that alliances among some
species of non-human primates are indeed of a transactional nature, with
both parties monitoring the contribution of the other and discontinuing the
collaboration if too large an imbalance is detected. De Waal (1992) reported
observing a subordinate male terminate his long-term alliance with an alpha
male in response to the alpha maIe’s increasingly frequent refusals to support him in contests with another male over access to oestrus females. Similarly, male Pupio unubis baboons who refuse to assist other males in
abducting females are less likely to receive aid than males who do (Alcock,
1984, p. 486). Woodruff and Premack (1979) reported that chimpanzees misinform or fail to inform individuals about the location of food if the individual failed to share food with them in the past.
Primate field studies, therefore, suggest that the capacity to detect and
respond appropriately to a variety of deontic structures and their violations
plays a crucial role in determining an individual’s fate within primate social
groups. Failure to adhere to these implicit prescriptive rules leads to banishment from the social group, a situation that can have disastrous consequences for survival. Failure to detect violations can result in encroachments
on one’s domination of resources. Clearly, if our reasoning architecture
evolved in response to the need to reason effectively about adaptively crucial
problems, and survival depends crucially on staying within the social group,
then few problems carry greater survival consequences among social species
than those involving deontic contents. This strongly suggests that deontic
reasoning strategies-or their precursors-are part of our primate genetic
heritage, and that violation detection is the most crucial of these strategies.
3. The Indicative-Deontic Distinction Emerges Early in Human
Development
Evidence that the indicative-deontic distinction constitutes a fundamental
division in the human reasoning architecture comes from a variety of
sources. The first and most incontrovertible evidence comes from my own
laboratory (Cummins, in press). I have found evidence of this distinction in
the reasoning performance of children as young as three years of age. The
procedure involves a story about a group of toy mice, some of whom are
inside a house and some of whom are in the backyard. Although the mice
are visually identical, some of them can squeak and some of them can’t. The
only way to tell the mice apart is to squeeze them. The drama in the story
comes from the fact that a neighbourhood cat chases the mice whenever he
hears anyone squeaking. For this reason, it is not safe for the squeaky mice
to be outside. In the indicative condition, Minnie Mouse tells the children
0 Blackwell Publishers Ltd. 1996
Deontic Reasoning Modules
171
that ’It’s not safe outside for the squeaky mouse, so all squeaky mice are in
the house‘. They are then asked which mice must be tested to see whether
Minnie is wrong, those that are in the house or those that are outside. In
the deontic condition, Queen Minnie Mouse tells them that ‘It’s not safe
outside for the squeaky mice, so all squeaky mice must stay in the house‘.
The child is then asked which mice must be tested to ensure that no one
disobeys the Queen, those that are in the house or those that are outside.
Notice that in both cases, the squeaky mice are in danger when outside. The
only difference is whether the child is testing a hypothesis (i.e. Is Minnie
Mouse wrong?) or ensuring that a deontic rule has not been broken (i.e. Did
anyone disobey the Queen?).
Across two experiments, 82% of four-year-olds and 64% of three-year-olds
in the deontic case chose to inspect the potentially violating case, that is,
those in the ‘backyard’. In contrast, only 32% of four-year-olds and 35% of
three-year-olds made that selection in the indicative case. The magnitude of
this difference is comparable to that observed in the adult reasoning literature (see Figure l).These results show quite clearly that even very young
children adopt of violation-detection strategy when reasoning about deontic
rules, and a confirmation-seeking strategy when reasoning about indicative
rules. Moreover, using a procedure similar to this with six- to nine-yearsolds, Girotto and his colleagues have demonstrated that the indicativdeontic distinction cannot be reduced to greater familiarity with individual rules
(Girotto, Gilly, Blaye and Light, 1989; Girotto, Light and Colbourn, 1988;
Light, Blaye, Gilly and Girotto, 1989; Light, Girotto and Legrenzi, 1990).This
distinction seems to emerge quite early in development, and persists
throughout adulthood. It is, it seems, a very fundamental cognitive distinction, and suggests the evocation of domain-specific reasoning strategies.
Recent work by Harris and NuAez (in press) suggests that children not
only distinguish between indicative and deontic rules, but that they genuinely understand deontic rules better than they do indicative rules. Their
task consisted of stories containing a rule in conditional form (that is, ’if p,
then q’), and four pictures that corresponded to each possible combination
of p and not-p with q and not-q (Experiment 4). In the deontic task, the rule
was a permission, such as ’If you play outside, you must wear a coat.’ The
pictures depicted e.g. a child playing outside wearing a coat (p and q), a
child playing indoors wearing a coat (not-p and q), a child playing indoors
without a coat (not-p and not-q), and a child playing outdoors without a
coat (p and not-q). The child’s task was to point to the picture showing the
story character ‘being naughty and not doing what he or she is supposed
to do‘. In the indicative task, the rule was an hypothesis, such as ’If Sally
plays outside, she always wears a coat.’ The pictures were the same, and
the child’s task was to point to the picture that showed the story character
’doing something different and not doing what she said’. Notice that this
task does not fall prey to the normative reasoning controversy cited above
because it simply measures one’s capacity to recognize rule violations,
0 Blackwell Publishers
Lfd. 1996
172
Mind b Language
regardless of what one believes about the necessity of seeking out such
violations to test the rules.
Harris and Nufiez found that three- and four-year-old children were better
at identifying the instances that violated the permission rule than the indicative rule. Moreover, they tended to justify their choices on the permission
task by referring to the fact that the protagonist had not met the condition
specified as necessary for taking the action. In contrast, they typically gave
irrelevant justifications (’It‘s just that one’) or were unable to justify their
choices on the indicative task. This suggests that they grasped the full meaning of the permission rules better than they did the indicative rules.
Perhaps the reason young humans show such precociousness for deontic
structures is because, like non-human primates, their social interactions
appear to be governed by dominance hierarchies which determine who is
permitted to play with whom where and with which toys. Transitive dominance hierarchies are evident in the interactions of children as young as
three, and can be reliably reported verbally by four-year-olds, meaning that
children as young as three can perform the transitive inferences that are
necessary to work out transitive dominance relations (Smith, 1988).Yet, this
skill does not transfer readily to non-social stimuli. Like non-human primates, they can perform object-based transitive reasoning only if they are
extensively drilled on the object pairs upon which the inference is to be
performed (Bryant and Trabasso, 1971).Truly content-free transitive reasoning does not reliably appear until six years of age (Smith, 1998, pp. 103-4).
Evidence of certain aspects of deontic reasoning is also apparent in the
social interactions of very young children. Within the first year of life, infants
participate in turn-taking games with adults, suggesting an appreciation of
the reciprocal nature of certain social interactions (Vandell and Wilson, 1987).
Reference to social rules appear in children’s justifications of their own
behaviour as early as 24 months of age (Dunn, 1988). And by the age of
three, children are selective in their distribution of altruistic acts, preferring
to aid those who have aided them in the past (Smith, 1988).
To summarize, the indicative-deontic distinction emerges early in human
development and colours the reasoning process throughout adulthood.
Regardless of their age, reasoners tend to adopt a violation-detection strategy
when reasoning about deontic rules, and a confirmation-seeking strategy
when reasoning about the epistemic status of indicative rules. This strongly
suggests a fundamental, primitive distinction in our cognitive architecture.
Furthermore, the structure of deontic situations is more easily grasped by
young children than is the structure of indicative rules. The ease and speed
with which young children learn about, detect, understand, and reason
about deontic situations is most consistent with the existence of an innate
domain-specific reasoning module that is evoked when a situation with
deontic content is encountered.
0 Blackwell Publishers Ltd. 1996
Deontic Reasoning Modules
173
4. The Indicative-Deontic Distinction in Cross-Cultural Studies of
Reasoning
If the indicative-deontic distinction reflects a fundamental division in our
reasoning architecture, then it should be observed universally, that is,
regardless of culture. Because folk societies are relatively isolated from modern industrialization and its education systems, they presumably provide a
more accurate picture of the types of reasoning strategies that characterized
our hunter-gatherer forebears. To my knowledge, no one has looked for content effects in preliterate cultures. Nonetheless, there is indirect evidence of
the indicative-deontic distinction in the reasoning of members of such
societies.
Indicative reasoning typically has been investigated in folk societies by
using classical syllogisms. An example cited by Scribner (1975, p. 155) on
syllogistic reasoning among the Kpelle of West Africa illustrates in high
relief the usual result observed in these studies:
E: All Kpelle men are rice farmers. Mr Smith (this is a Western
name) is not a rice farmer. Is he a Kpelle man?
S: I don’t know the man in person. I have not laid eyes on the
man himself.
E: Just think about the statement.
S: If I know him in person, I can answer that question, but since I
do not know him in person I cannot answer that question.
E: Try and answer from your Kpelle sense.
S: If you know a person, if a question comes u p about him, you are
able to answer. But if you do not know a person, if a question
comes up about him, it’s hard for you to answer it.’
As this excerpt shows, the reasoner does not seem to grasp the fact that
the dilemma can be resolved via reasoning. Instead, he is concerned with
retrieving and verifying facts, a strategy that does not fare well in this type
of task. In fact, the reasoner’s responses seem to suggest that he believes
he is trying to explain something very simple to an experimenter who is a
complete dunce.
A very different picture emerges, however, when members of preliterate
cultures are asked to reason about moral dilemmas, problems with clear
deontic contents. Comparison of studies from eight different traditional folk
societies showed that 100% of the samples reported reasoning that was
dominated by concern with maintaining social equilibrium and social harmony through the adherence to social regulations (Snarey, 1985). Like the
societies of non-human primates, social accord in such societies is maintained through what Pope Edwards described as ’cooperation and reciprocal
obligations within a hierarchical structure of authority relationships’ (Pope
Edwards, 1982, p. 276). In other words, a dominance hierarchy.
One oft-cited study by Harkness, Pope Edwards and Super (1981) on rural
0 Blackwell Publishers Ltd. 1996
174
Mind & Language
Kenyans (Kipsigis community) makes this clear. The reasoning of six traditional leaders on several moral dilemma reasoning tasks showed three consistent results. First, in contrast to content-free reasoning, deontic reasoning
is remarkably sophisticated. For example, contrast the following protocol
with the one reported above. The dilemma was whether a boy should obey
his father and give him money he had earned when the father had promised
the boy he could keep it for himself (Harkness, Pope Edwards and Super,
1981, pp. 595-6):
A child has to give you what you ask for just in the same way as
when he asks for anything you give it to him. Why then should he
be selfish with what he has? A parent loves his child and maybe
[the son] refused without knowing the need of helping his father
. . . By showing respect to one another, friendship between us is
assured and as a result this will increase the prosperity of our family.
Unlike the syllogism protocol, the reasoner does not complain that he does
not know the people in question, and he does not attempt to solve the
dilemma through fact retrieval. Instead, this protocol reflects a weaIth of
deontic concepts, particularly reciprocal obligation. Reciprocity was in fact
a recurring theme in many of the observed protocols, although, interestingly,
not in terms of a strict accounting of benefits given or received. These men
believed that when an item was given in good faith, the receiver was honourbound to return the favour in the future. (In fact, so strong and pervasive
is this belief and so freely are goods given away that western missionaries
often mistakenly concluded that members of the pre-literate cultures had no
concept of private property.) This type of reciprocity is evidence of sensitivity to obligation structures-one is obligated to return the help g'wen one
in time of need.
Second, the dominance hierarchy that pervades social interactions and
social reasoning in non-human primate societies and children's social groups
had its counterpart here in that all six men settled dilemmas in favour of
the person with higher rank or status.
Third, a recurring issue in these protocols is the treatment of those who
break social rules. The traditional Kipsigis response was banishment, that is,
exclusion from future social interactions and social contracts. Clearly, the capacity
to detect violations of deontic rules looms large in folk societies, where the
fate of each individual depends crucially on the cooperative acceptance of
social norms.
To summarize, cross-cultural studies of human reasoning show a clear
distinction between indicative reasoning and deontic reasoning. In contrast
to the sparse (or absent) reasoning strategies evoked by indicative reasoning
tasks, deontic reasoning tasks evoke strategies that are sophisticated and
conceptually rich. Moreover, the strategy of choice in deontic situations was
violation-detection. Clearly, the evocation of violation-detection reasoning
0 Blackwell Publishers Ltd. 1996
Deontic Reasoning Modules
175
strategies in response to deontic rules does not depend on a reasoner’s culture nor on the level of education achieved. Just as clearly, two recurring
themes in these reasoning protocols are authority (dominance) and reciprocal obligations. This is particularly crucial to my position because the structure of these societies (particularly hunter-gatherer societies) and the survival pressures their members face provide a glimpse into our own
evolutionary past. The need to reason effectively about authority/dominance
structures and reciprocal obligations looms exceedingly large in these communities. Failure to adhere to permission and obligation structures Ieads
to banishment from the social group, a situation that can have disastrous
consequences for survival. Clearly, if our reasoning architecture evolved in
response to the need to reason effectively about adaptively crucial problems,
and survival depends crucially on staying within the social group, then few
problems carry greater survival consequences in hunter-gatherer societies
than those involving deontic contents.
5. The Neurological Dissociability of Indicative and Deontic
Reasoning
Prefrontal lobe syndrome is a pattern of impaired reasoning performance
that results from bilateral damage to the ventromedial prefrontal cortical
lobes (Damasio, 1994). This syndrome is characterized by an impaired
capacity to reason effectively about socio/emotional stimuli while leaving
other types of intelligent reasoning virtually untouched. Damasio (1994)
reported the case of a successful, middle-aged businessman who suffered
bilateral damage to the prefrontal cortex as a result of a benign tumour.
After surgery to remove the tumour, the patient regained all of his previous
intellectual capacities, yet went on to make disastrous personal and financial
decisions which resulted in two divorces and a bankruptcy. Closer inspection of his reasoning capacities in the laboratory revealed a selective impairment socio-emotional reasoning.
Ablation studies in monkeys provide even more striking evidence of specific social reasoning impairments (Damasio, 1994, pp. 74-5). Monkeys with
bilateral prefrontal ablations (both ventromedial and dorsolateral) do not
maintain normal social relations within their troops despite the fact that
nothing in their physical appearance has changed. They show diminished
self-grooming and reciprocal grooming behaviour, greatly reduced affective
interactions with others, diminished facial expressions and vocalizations,
and sexual indifference. They can no longer relate properly to others in their
troop and others cannot relate to them. As a result, and this is crucial to my
position, they can no longer operate effectively within their social dominance
hierarchies, or as Damasio puts it (p. 75):
It is fair to assume that monkeys with prefrontal damage can no
longer follow the complex social conventions characteristic of the
0 Blackwell Publishers Ltd. 1996
176
Mind 6 Language
organization of a monkey troop (hierarchicalrelations of its different
members, dominance of certain females and males over other members, and so on.)
Damage to other sections of the cortex-ven
those resulting in paralysisdo not impair these social skills.
Prefrontal syndrome is the ’smoking gun’ in my collection of evidence
since it constitutes definitive evidence that social reasoning can be selectively
dissociated from other types of reasoning at the neurological level, and that
damage to these areas impacts most severely on the capacity to respond
effectively to the social rules that underlie the dominance hierarchy.
Before leaving this section, I would like to contrast prefrontal syndrome
with another neurological syndrome in order to make an important point.
Autism is a neurodevelopmental syndrome whose most vivid impact at the
cognitive level is an impaired ability to reason about the mental states of
others (see Baron-Cohen, Tager-Flusberg and Cohen, 1993). As one autistic
adult put it ’Other people seem to have a special sense by which they can
read other people’s thoughts’ (Frith, Morton and Leslie, 1991, p. 436b-a
description that bears an uncanny resemblance to Krebs and Dawkins’ (1984)
notion of ‘mind-reading’, that is, the capacity to forecast the behaviour of
others. In autism, this ’mind blindness’ (Baron-Cohen, 1995) makes it
extremely difficult (if not impossible in some cases) to engage in normal
reciprocal social interactions. Moreover, autistic individuals typically fail
reasoning tasks that require them to reason effectively about others’ beliefs
(Frith et al., 1991).This suggests that ’theory of mind’ reasoning is neurologically distinct from other types of reasoning.
’Theory of mind‘ reasoning, however, should not be confused with deontic
reasoning. Theoretically, the capacity to engage in deontic reasoning does
not presuppose a capacity to represent or reason about others’ mental states.
Put simply, it is conceivable that one can appreciate which actions are permitted or forbidden in which circumstances without also appreciating the
motivation or mental states of the individual making the rules. For example,
even three-year-olds reason about deontic rules in the same way adults do
(Cummins, in press; Harris and Nuiiez, in press), but a developmental
change in performance occurs between three and four years of age in terms
of the effectiveness with which children can reason about others’ beliefs
(Baron-Cohen, Leslie and Frith, 1985; Wimmer and Perner, 1983). This suggests that the capacity to engage in deontic reasoning may emerge earlier
(and perhaps be a more fundamental distinction in mammalian reasoning
architectures) than the capacity to reason effectively about the belief states
of others.
6. Alternative Explanations of the Indicative-Deontic Distinction
The indicativedeontic distinction in human reasoning performance is a real
distinction that emerges early in human development, persists into adultQ Blackwell Publishers Ltd. 1996
Deontic Reasoning Modules
177
hood, is observed among members of pre-literate societies, and has its
counterpart in the reasoning of non-human primates. When reasoning about
deontic rules, reasoners tend to adopt a violation-detection strategy, a strategy that they do not adopt when reasoning about the epistemic status of
indicative rules. I have argued that the pervasiveness and robustness of this
effect is due to the evocation of innate deontic reasoning modules that
evolved to enable social species to construct and exploit the deontic structures inherent in the dominance hierarchy-the social structure that determines overall survival and reproductive success. Furthermore, I have shown
that damage to particular areas of the brain selectively impairs the capacity
to engage in social reasoning, particularly reasoning about the dominance
hierarchy.
In contrast to this view, four distinct points of view have been advanced
to explain content effects in general and the indicative-deontic distinction
in particular. The first, championed by Braine and OBrien, Rips, and Osherson, is that human reasoning is based on the activation of syntax-sensitive
rules, and content effects such as these reflect the operation of factors outside
the rule base itself (Braine, 1978; Braine and OBrien, 1991; Braine, Reiser
and Rumain, 1984; Rumain, Connell and Braine, 1983; Osherson, 1974, 1975;
Rips, 1983, 1994). The major difficulty with this view is the robust and pervasive effect of content on human reasoning performance. In order to
account for observed performance patterns, major proponents of these views
have taken to incorporating content-sensitive parameters (Braine and
OBrien, 1991) or modal operators (Rips, 1983, 1994) into their reasoning
systems. But this essentially concedes the point that the human reasoning
architecture must consist of something more than syntax-sensitive rules if
the robustness and pervasiveness of content effects are to be explained.
The second view, proposed by Cheng, Holyoak and their colleagues, is
that humans reason by activating domain-specific bodies of rules, called
schemas, which are acquired during the lifetime of the individual through
general inductive mechanisms (Cheng and Holyoak, 1985, 1989; Cheng,
Holyoak, Nisbett and Oliver, 1986). Content effects like these are therefore
a natural reflection of the types of schemas so induced. According to the
schema theory, people adopt different strategies when reasoning about the
'travel' problem example and the 'transportation law' problem described
above because the latter is a permission, a type of social problem for which
most people have induced a particular schema. What we understand about
permission situations is codified in a schema consisting of four rules, illustrated in Figure 2(a). A schema has also been proposed for obligations, and
is illustrated in Figure 2(b).
One objection to Cheng and Holyoaks pragmatic schema view as it is
currently stated is that its strictly empiricist stance does not sort well with
the early emergence of deontic reasoning during childhood relative to other
types of reasoning and the evidence of neurological substrates devoted to
this type of social reasoning. Observing reasoning effects among three-yearolds that are as large as those cited above is generally taken as evidence in
0 Blackwell Publishers Ltd. 1996
178
Mind t? Language
(a)
Permission Schema proposed by Cheng and Holyoak (2985, p. 397)
Rule 1: If the action is to be taken, then the precondition must be satisfied.
Rule 2: If the action is NOT to be taken, then the precondition need NOT be satisfied.
Rule 3: If the precondition is satisfied, then the action may be taken.
Rule 4: If the precondition is NOT satisfied, then the action must NOT be taken.
(b)
Obligation Schema proposed by Cheng, Holyoak, Nisbett, and Oliver (1986, pp. 325-326)
Rule 1: If initial situation I occurs, then C must be done.
Rule 2: If situation I does NOT occur, C need NOT be done.
Rule 3: If C is done, then it does NOT matter whether I occurred; the obligation
cannot be violated in this case.
Rule 4: If C has NOT been done, then I must NOT have occurred.
Note: In each schema, Rules 1 and 4 describe situations in which a violation of the
regulation is possible.
Figure 2 Permission (a) and obligation (b) reasoning schemes proposed by
Cheng and her colleagues
the developmental literature of innate predispositions to interpret certain
stimuli in particular ways. For example, the ease with which children in this
age-group exploit ontological categories in order to discipline their inductive
generalizations is taken as evidence of innate or early emerging knowledge
concerning these categories (e.g. Carey, 1985; Gelman and Markman, 1985,
1986).PRS would sort better with the data if it were couched in terms of an
innate or early emerging predisposition (or ’preparedness’; Seligman, 1970)
to attend to the structure of deontic situations.
The same objection can be raised concerning the third view, proposed by
Manktelow and Over (1990, 1991, 19951, which attributes the preference for
violation-detecting strategies on problems with deontic contents to the construction of mental models based on social roles and subjective utility. They
argue that deontic reasoning is psychologically distinct from indicative
reasoning. In the former, one is concerned about what one may or must do,
and hence one’s reasoning is focused on evaluating the subjective utility of
the possible outcomes of one’s actions. In the latter case, one is concerned
primarily with a factual state of affairs, and hence one’s reasoning is focused
on evaluating the truth of sentences describing these states. Their objection
to Cheng’s schema view is that it provides no way of assessing utilities of
possible actions, and no way of capturing the fact that subjective utilities
differ depending on which perspective one adopts during reasoning. For
example, consider the rule ‘If you tidy your room, then you may go outside
to play.’ A child would place greater utility on going outside to play rather
than being kept in, and hence his or her reasoning would focus on these
0 BIackwell Publishers Ltd. 1996
Deontic Reasoning Modules
179
possible outcomes when determining whether to take the action of tidying
the room. The parent uttering the statement, on the other hand, places greater utility on having tidy rooms than untidy rooms, and hence would focus
on these outcomes when determining whether or not to allow the child to go
outside to play. The authors report a series of experiments in which deontic
reasoning performance on the Wason card selection task varied as a function
of perspective taken and the subjective utility assigned to possible outcomes
(Manktelow and Over, 1991).
Unlike the general induction view offered by Cheng and Holyoak, Manktelow and Over’s analysis of denotic reasoning clearly indicates that deontic
reasoning is distinct from other types of reasoning. It is neutral, however,
with respect to the innateness argument. Presumably, one places greater utility on actions that enhance one’s survival than on those that jeopardize it
whether these actions are learned (e.g. ’don’t drink the water’) or innate
(e.g., ’fight or flee’). If I am right that deontic reasoning strategies are innate,
then their analysis constitutes a possible description of the components of
these innate strategies. As such, their view is not a competitor for the view
offered here as much as a proposal concerning the components (subjective
utility) and strategies (maximizing subjective utility) a deontic reasoning
module might contain.
The same could be said of the rational analysis offered by Oaksford and
Chater (1994).These authors offer an elegant analysis of performance on the
indicative and deontic versions of the card selection task based on information theory (Shannon and Weaver, 1949; Wiener, 19481, Bayesian decisionmaking, and subjective utility. Indicative rule-testing is modelled as hypothesis-testing under conditions of uncertainty where the reasoner chooses
experiments based on the amount of information that the experiment is
expected to provide. Rational decision-making is defined as choosing to conduct experiments (card-turnings on the selection task) that are expected to
lead to the greatest reduction in uncertainty (greatest information gain) concerning which of two hypotheses is true (i.e., ’if p is true, then q must be
true as well’ or ’p and q are independent’.) When it can be assumed that
the incidences of p and q are low in the sample space, then the ordering of
information gain for each of the four standard cards in the selection task is
p > q > not-q > not-p, explaining why reasoners prefer to turn over the q
card rather than the not-q card. As the probabilities associated with the incidences of p and q increase, so does the information value associated with
selecting the not-q card. In contrast, selection performance on deontic versions of the selection task is explained using the same probabilistic model
coupled with the assignment of different subjective utilities to each card
depending on the viewpoint that is adopted during reasoning. Rational
decision-making is defined in this case as selecting cards that maximize
expected utility. This has the effect of producing a violation-detection strategy because the model predicts that violating instances have the greatest
expected utility. Domain-dependent changes in selection performance therefore result from domain-specific knowledge that influences the parameters
Q Blackwell Publishers Ltd. 1996
180
Mind & Language
in the model. This analysis constitutes a proposal of how domain-specific
strategies are instantiated in our reasoning architecture, or, to put it more
simply, what might be ‘inside’ the deontic-reasoning module.
The fifth view, proposed by Cosmides and Tooby, attributes the deontic
effect to the activation of innate reasoning algorithms, algorithms that were
shaped by evolutionary forces in order to facilitate reasoning about social
exchange (Cosmides, 1989; Cosmides and Tooby, 1989, 1992, 1994). They
define social exchange as a species of reciprocal altruism, that is, the
cooperation of two or more individuals for mutual benefit. Using this definition, social exchange is a subset of deontic situations, concerning those
situations that involve reciprocal obligations. Cosmides and Tooby argue
that the need for engaging in exchange of goods and services loomed large
among our Pleistocene hunter-gatherer ancestors. Crucial to this theory is
the notion of cheater detection. Modelling research based on game theory
has repeatedly shown that reciprocal altruism can emerge as an evolutionarily stable strategy only if the participants are capable of recognizing
individuals so that those who cheat may be excluded from future transactions (Axelrod, 1984; Axelrod and Hamilton, 1981; Maynard Smith, 1982;
Trivers, 1971). They define cheating as taking a benefit without paying a
cost. According to socia1 exchange theory, people typically fail hypothesis
testing problems (such as the ’travel’ problem above) and solve denotic problems (such as the ’transportation law’ problem) because the latter trigger a
cheater detection strategy while the former do not.
There are two difficulties with this view. The first is an empirical one,
namely that robust deontic content effects can be observed regardless of the
cost associated with a particular action. For example, Cheng and Holyoak
(1989) report an experiment in which reasoners were presented the rule ’If
you go out at night, then you must tie a piece of volcanic rock around your
ankle’. A preference for violation-detection was observed with this conditional relative to an indicative conditional regardless of whether the volcanic rock was abundant and free or expensive and difficult to obtain.
The second difficulty is that, as I pointed out in this paper, deontic problems are ones that are faced by any primate species, and indeed, any social
species. This strongly suggests that the deontic content effect reaches far
more deeply into our evolutionary past than the Pleistocene era, and underlies efficient reasoning about deontic contents other than the reciprocal obligation structure that characterizes the exchange of goods and services. From
my perspective, this means that a primitive reasoning module governing
simple yet pervasive deontic situations emerged first during mammalian
evolution, and constituted the foundation upon which our species’ advanced
capacity for complex social exchange evolved.
To summarize, theories that describe human reasoning in terms of the
activation of syntactically-drivenrules do not adequately explain the pervasiveness and robustness of content effects. The strictly empiricist version of
pragmatic schema theory proposed by Cheng and Holyoak and the decisionmaking models proposed by Manktelow and Over and by Oaksford and
0 Blackwell Publishers Ltd. 1996
Deonfic Reasoning Modules
181
Chater do not adequately explain why the structure of deontic situations is
so readily understandable to even young children whereas the structure of
other, equally familiar situations are not, nor why social reasoning appears
to be dissociable at the neurological level. Social exchange theory runs
aground on deontic effects that are not readily amenable to cost-benefit
analyses. In contrast, the view offered here provides an explanation for each
of these observations. The need to construct, maintain, and exploit dominance hierarchies constituted strong pressure favouring the evolution of a
cognitive architecture capable of effective deontic reasoning. Because this
architecture is part of our genetic heritage, effective deontic reasoning
emerges early in development, is readily observed in adult humans regardless of culture, is readily observed among non-human primates, and is part
of the social reasoning that is dissociable at the neurological level.
7. lmplications for a Theory of Human Reasoning
The final question to be addressed is what implication the indicative-deontic
distinction has for theories of human reasoning. Content effects in general
have been interpreted to mean that the human reasoning architecture is
imbued with domain-specific characteristics.The movement toward domainspecificity in the adult reasoning literature is consistent with a burgeoning
developmental literature showing innate or early-emerging domain-specific
knowledge. During the first year of life, infants evidence an appreciation of
the distinction between living and non-living things, an appreciation that is
marked by distinct sets of domain-specific knowledge. They appreciate that
objects are solid, rigid, permanent entities that cannot move of their own
volition, can causally influence each other only through direct contact, and
whose movements are continuous in space and time (Baillargeon, 1987; Leslie, 1984; Leslie and Keeble, 1987; Spelke, 1994). In contrast, infants in this
same age group access a different set of knowledge concerning social stimuli,
knowledge that centers on the intentional and reciprocal nature of social
interactions. They become upset when a person stands before them perfectly
still, but show no distress when a similarly sized stationary object is placed
before them (Tronick, Als, Adamson, Wise and Brazelton, 1978). They
engage in reciprocal, turn-taking play with others (Vandell and Wilson,
19871, and respond appropriately to the meaning conveyed by a variety of
emotional facial expressions (Campos and Stenberg, 1981). As early as the
third year of life, children distinguish between physical and social causation,
attributing physical events to energy transmissions between causes and
effects, while attributing social events to the enactment of the intentions of
the parties involved (Shulz, 1982).They also discipline their inductive generalizations to respect broad ontological category membership, such as the
distinction between natural kinds and artifacts, and animate and inanimate
objects (Gelman and Markman, 1986, 1987; Carey, 1985; Keil, 1994). The rapidity and ease with which children make distinctions between social and
0 Blackwell Publishers Ltd 1996
182
Mind G. Language
physical stimuli suggests a natural division in the innate nature of the mind,
or, to put it in more contemporary terms, the innate structure of the human
cognitive architecture.
But does this mean that all human reasoning can be attributed to the workings of separate, domain-specific modules? I don’t think so. Intelligence
researchers have compiled an impressive body of evidence for ’g’, or a general intelligence factor that differs among species and among members of
any given species. Humans are more intelligent across the board than chimpanzees, and individual humans and chimpanzees differ among themselves
in terms of overall intelligence. In short, one could imagine special-purpose
modules evolving within a context of greater overall reasoning capacity that
resulted from brain expansion and specialization for language. This seems
particularly plausible when one considers the enormous reasoning advantages that arose from the evolution of natural language in Homo sapiens, the
sine qua non of an abstract, symbolic reasoning and communication system. It
would be astonishing indeed if the capacity to exploit syntactic features of
arguments were not part of the human reasoning architecture.
As an example, consider that children as young as two and a half years
of age utter conditional statements to describe hypothetical events in a variety of domains (Bloom, Lahey, Hood, Lifter and Feiss, 1980; Bowerman,
1986; Reilly, 1986).Young children (and adults) also readily draw the Modus
Ponens inference, even with relatively content-free conditionals, such as ’If
there is an A, then there is a B (Braine, 1978; Braine, Reiser and Rumain,
1984; Osherson, 1974,1975; Rumain, Connell and Braine, 1983).This suggests
that at least part of our conditional reasoning strategies are driven by purely
syntactic considerations. But when the constituent propositions of a conditional utterance express a deontic structure, a specific violation-detection
reasoning module is triggered-even among humans as young as three years
old. This suggests that the syntactic reasoning process can be overridden
(or supplemented) by the evocation of a domain-specific deontic reasoning
module wherein one focuses on violation detection, and perhaps subjective utility.
Alternatively, Cosmides and Tooby (1994) recently suggested that human
reasoning performance might be governed by a Universal Grammar of social
interactions. Detecting cheaters in this grammar is analogous to detecting
utterances that are not well formed, an analogy also suggested by Much and
Schweder in 1978. The idea is that utterances such as ‘I want to help him
because whenever I’m in trouble he refuses to help me’ are not well formed
because they do not describe inferences that members of a social community
would judge to be sensible.
It is not clear why Cosmides and Tooby decided to appeal to a formalism
that was developed to capture the rules of well-formedness (namely a
grammar) when a type of formalism already exists that was developed to
capture the rules of reasonable inference (namely a logic). There are perhaps
two motivations for this choice. The first is that, since the Chomskian revolution in linguistics, it has become customary to think of language and all its
0 Blackwell Publishers Ltd. 1996
Deontic Reasoning Modules
183
accoutrements as biological in origin, and perhaps this makes it easier to
conceive of a grammar-type formalism having evolved as opposed to a logiclike formalism. But there is no a priori reason to believe that evolved reasoning modules cannot consist of rules of inference rather than rules of grammar.
The second motivation stems from an objection explicitly raised by
Cosmides and Tooby in relation to content-free, rule-based reasoners. Automated reasoners of this type typically fall prey to the frame problem, that is,
they have difficulty determining which information to attend to and which
inference paths are fruitful ones to follow. The standard solution is to constrain the reasoning space to a particular domain (e.g. Freuder and Mackworth, 1994). Medical reasoners constrain their inferences only to medical
diagnosis, computer system configuration reasoners reason only about system configuration, and geology reasoners reason only about geology. By constraining the types of inputs a reasoner receives and the types of inferences
the system can make, the frame problem is greatly reduced.
Which brings us back again to the notion of domain specificity in human
reasoning. Given that even rule-based theories have been modified to reflect
the fact that certain contents produce robust qualitative shifts in reasoning
performance, it is difficult to avoid concluding that there is indeed something domain-specific about significant segments of our reasoning architecture. The robustness of the deontic effect in adult reasoning regardless of
culture, its early emergence in human development, and the evolutionary
pressure to select for this type of reasoning given the central role played
by deontic reasoning problems in primate dominance hierarchies provide a
compelling case that at least one of these domain-specific segments is
devoted to deontic reasoning.
Cognitive Science
University of Arizona
References
Aicock, J. 1984: Animal Behavior: A n Evolutionary Approach. 4th edition. Sunderland, MA.: Sinauer Associates.
Aruguete, M. 1994: Cognition, Tradition and the Explanation of Social Behavior
in Non-human Primates. Review of Social Processes and Mental Abilities in
Non-human Primates. American Journal of Primatology, 33, 71-4.
Axelrod, R. 1984: The Evolution of Cooperation. New York: Basic Books.
Axelrod, R. and Hamilton, W. D. 1981: The Evolution of Cooperation. Science,
211, 1390-96.
Baillargeon,R. 1987 Object Permanence in 31- and 44-month-old Infants. Developmental Psychology, 23, 655-64.
Baron-Cohen, S. 1995: Mindblindness: An Essay on Autism and Theory of Mind.
Cambridge, MA.: MIT Press.
Baron-Cohen, S., Leslie, A.M. and Frith, U. 1985: Does the Autistic Child Have
a 'Theory of Mind'? Cognition, 21, 37-46.
0 Blackwell Publishers Ltd. 1996
184
Mind 6 Language
Baron-Cohen, S., Tager-Flusberg, H. and Cohen, D. 1993: Understanding Other
Minds: Perspectives From Autism. Oxford University Press.
Bloom, L., Lahey, M., Hood, L., Lifter, K. and Feiss, K. 1980: Complex sentences:
Acquisition of Syntactic Connectives and the Semantic Relations They
Encode. Journal of Child Language, 7, 235-61.
Bowerman, M. 1986: First Steps in Acquiring Conditionals. In E. Traugott, A.
ter Meulen, J.S. Reilly and C.A. Ferguson (eds), On Conditionals. Cambridge
University Press.
Braine, M.D.S. 1978: On the Relation Between the Natural Logic of Reasoning
and Standard Logic. Psychological Review, 85, 1-21.
Braine, M.D.S. and OBrien, D.P. 1991: A Theory of If: A Lexical Entry, Reasoning
Program, and Pragmatic Principles. Psychological Review, 98, 182-203.
Braine, M.D.S., Reiser, B.J. and Rumain, B. 1984: Some Empirical Justification for
a Theory of Natural Propositional Logic. In G.H. Bower (ed), The Psychology
of Learning and Motivation, Volume 18. New York: Academic Press.
Bryant, P.E. and Trabasso, T. 1971: Transitive Inference and Memory in Young
Children. Nature, 240, 456-8.
Byrne, R.M.J. 1989: Suppressing Valid Inferences with Conditionals. Cognition,
31. 61-83.
Byrne,'R.W. and Whiten, A. 1988: Tactical Deception of Familiar Individuals in
Baboons. In R.W. Byme and A. Whiten (eds), Machiavellian Intelligence: Social
Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford
University Press.
Campos, J.J. and Stenberg, C. 1981: Perception, Appraisal, and Emotion: the
Onset of Social Referencing. In M. Lewis and L. Rosenblum (eds), Infant
Social Cognition: Empirical and Theoretical Considerations. Hillsdale, NJ:
Erlbaum.
Carey, 1985: Conceptual change in childhood. Cambridge, MA.: MIT Press.
Chapais, B. 1988: Rank Maintenance in Female Japanese Macaques: Experimental
Evidence for Social Dependency. BehazTiour, 104, 41-59.
Chapais, B. 1992: Role of Alliances in the Social Inheritance of Rank Among
Female Primates. In A. Harcourt and F.B.M. de Waal (eds), Cooperation in
Contests in Animals and Humans. Oxford University Press.
Cheney, D.L. 1978: Interactions of Immature Male and Female Baboons With
Adult Females. Animal Behavior, 26, 389-408.
Cheney, D.L. and Seyfarth, R.M. 1985: The Social and Non-Social World of NonHuman Primates. In R.A. Hinde, A-N. Perret-Clermont and J. StevensonHinde (eds), Social Relationships and Cognitive Development. Oxford: Clarendon Press.
Cheney, D.L. and Seyfarth, R.M. 1988: Social and Non-Social Knowledge in Vervet Monkeys. In R.W. Byrne and A. Whiten (eds),Machiavellian Intelligence:
Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans.
Oxford University Press.
Cheney, D.L. and Seyfarth, R.M. 1990: How Monkeys See The World. Chicago University Press.
Cheng, P.W. and Holyoak, K.J. 1985: Pragmatic Reasoning Schemas. Cognitive
Psychology, 17, 391416.
Cheng, P.W. and Holyoak, K.J. 1989: On the Natural Selection of Reasoning Theories. Cognition, 33, 285-313.
Cheng, P.W., Holyoak, K.J., Nisbett, R.E. and Oliver, L.M. 1986 Pragmatic vs
0 B I a c h e l l Publishers Ltd. 1996
Deontic Reasoning Modules
185
Syntactic Approaches to Training Deductive Reasoning. Cognitive Psychology,
18. 293-328.
Churchland, P.S. 1986: Neurophilosophy. Cambridge, MA.: MIT Press.
Clutton-Brock, T.H. and Harvey, P.H. 1976: Evolutionary Rules and Primate
Societies. In P.P.G. Bateson and R.A. Hinde (eds), Growing Points in Ethology.
Cambridge University Press.
Cosmides, L. 1989: The Logic of Social Exchange: Has Natural Selection Shaped
How Humans Reason? Studies with the Wason Selection Task. Cognition,
31, 187-276.
Cosmides, L. and Tooby, J. 1989: Evolutionary Psychology and the Generation
Culture, Part 11. Case study: A Computational Theory of Social Exchange.
Ethology and Sociobiology, 10, 51-97.
Cosmides, L. and Tooby, J. 1992: Cognitive Adaptations for Social Exchange. In
J. Barkow, L. Cosmides and J. Tooby (eds), The Adapted Mind: Evolutionay
Psychology and the Generation of Culture. New York: Oxford University Press.
Cosmides, L. and Tooby, J. 1994: Beyond Intuition and Instinct Blindness:
Toward an Evolutionarily Rigorous Cognitive Science. Cognition, 50, 41-77.
Cox, J.R. and Griggs, R.A. 1982: The Effects of Experience on Performance in
Wason's selection task. Memoy and Cognition, 10, 496-502.
Cummins, D.D. 1995: Naive Theories and Causal Deduction. M e m o y and Cognition, 23, 646-58.
Cummins, D.D. in press: Evidence of Deontic Reasoning in 3- and 4-year-old
Children. M e m o y and Cognition.
Cummins, D.D., Lubart, T., Alsknis, 0. and Rist, R. 1991: Conditional Reasoning
and Causation. Memory and Cognition, 19, 274-82.
Damasio, A.R. 1994: Descartes' Error: Emotion, Reason, and the Human Brain. New
York Grosset/Putnam.
Dasser, V. 1985: Cognitive Complexity in Primate Social Relationships. In R.A.
Hinde, A.N. Perret-Clemont and J. Stevenson-Hinde (eds), Social Relationships and Cognitive Development. Oxford: Clarendon Press.
Datta, S.B. 1983a: Relative Power and the Acquisition of Rank. In R.A. Hinde
(ed.), Primate Social Relationships. Oxford: Basil Blackwell.
Datta, S.B. 1983b: Relative Power and the Maintenance of Rank. In R.A. Hinde
(ed), Primate Social Relationships. Oxford: Basil BlackweIl.
Datta, S.B. 1983c: Patterns of Agonistic Interference. In R.A. Hinde (ed.), Primate
Social Relationships. Oxford: Basil Blackwell.
De Waal, H. 1982: Chimpanzee Politics. Baltimore: Johns-Hopkins University
Press.
De Waal, H. 1988: Chimpanzee Politics. In R.W. Byrne and A. Whiten (eds),
Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford University Press.
De Waal, H. 1992: Coalitions as Part of Reciprocal Relations in the Arnhem Chimpanzee Colony. In Harcourt, A.H. and de Waal, F.B.M. (eds) 1992: Coalitions
and Alliances in Humans and Other Animals. Oxford University Press.
Evans, J. St. B.T. 1989: Bias in Human Reasoning. Hillsdale, N.J.: Erlbaum.
Federov, V.V. 1972: Theory of Optimal Experiments. San Diego, CA.: Academic
Press.
Fillenbaum, S. 1975: If: Some Uses. Psychological Research, 37, 245-60.
Fillenbaum, S. 1976: Inducements: On Phrasing and Logic of Conditional Promises, Threats, and Warnings. Psychological Research, 38, 231-50.
0 Blackwell Publishers Ltd. 1996
186
Mind & Language
Fillenbaum, S. 1978: How to Do Some Things with IF. In J.W. Cotton and R.L.
Klatzky (eds), Semantic Factors in Cognition. Hillsdale, N.J.: Erlbaum.
Freuder, E.C. and Mackworth, A.K. 1994: Constraint-Based Reasoning. Cambridge,
MA.: MIT Press.
Frith, U., Morton, J. and Leslie, A.M. 1991: The Cognitive Basis of a Biological
Disorder: Autism. Trends in Neuroscience, 14, 433-38.
Gelman, S.A. and Markman, E.M. 1986: Categories and Induction in Young Children. Cognition, 23, 183-209.
Gelman, S.A. and Markman, E.M. 1987: Young, Children’s Inductions from Natural Kinds: The Role of Categories and Kppearances. Child Development, 58,
15324.
Gigerenzer, G. 1993a:How to Make Cognitive Illusions Disappear: Beyond ’Heuristics and Biases’. European Review of Social Psychology, Volume 2. New
York Wiley.
Gigerenzer, G. 1993b The Bounded Rationality of Probabilistic Mental Models.
In K.I. Manktelow and D.E. Over (eds), Rationality. London: Routledge.
Gigerenzer, G. in press: Rationality: Why Social Context Matters. In P.B. Bates
and U. Staudinger (eds), Interactive Minds: Life-Span Perspectives on the Social
Foundation of Cognition. Cambridge University Press.
Gigerenzer, G. and Hoffrage, U. in press: How to Improve Bayesian Reasoning
Without Instruction: Frequency Formats. Psychological Review.
Gigerenzer, G. and Hug, K. 1992: Domain-specific Reasoning: Social Contracts,
Cheating, and Perspective Change. Cognition, 43, 127-71.
Gilhooly, K.J. and Falconer, W.A. 1974: Concrete and Abstract Terms and
Relations in Testing a Rule. Quarterly Journal of Experimental Psychology, 26,
355-59.
Gillan, D.J. 1981: Reasoning in the Chimpanzee: 11. Transitive Inference. Journal
of Experimental Psychology: Animal Behavioural Processes, 7 150-64.
Girotto, V., Gilly, M., Blaye, A. and Light, P. 1989: Children’s Performance in the
selection task: Plausibility and Familiarity. British Journal of Psychology, 80,
79-95.
Girotto, V., Light, P. and Colbourn, C.J. 1988: Pragmatic Schemas and Conditional Reasoning in Children. Quarterly Journal of Experimental Psychology:
Human Experimental Psychology, 40, 469-82.
Goodall, J. 1986: The Chimpanzees of Gombe. Cambridge: Belknap Press.
Griggs, R.A. and Cox, J.R. 1982: The Elusive Thematic-materials Effect in Wason’s
selection task. British Journal of Psychology, 73, 407-20.
Griggs, R.A. and Cox, J.R. 1983: The Effects of Problem Content and Negation
on Wason‘s Selection Task. Quarterly Journal of Experimental Psychology, 35A,
519-33.
Harcourt A.H. 1988: Alliances in Contests and Social Intelligence. In R.W. Byrne
and A. Whiten (eds), Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford University Press.
Harcourt, A.H. and Stewart, K.J. 1987: The Influence of Help in Contests on
Dominance Rank in Primates: Hints from gorillas. Animal Behauiour, 35,
182-90.
Harcourt, A.H. and de Waal, F.B.M. 1992: Coalitions and Alliances in Humans and
3 h e r Animals. Oxford University Press.
Harkness, S., Pope Edwards, C. and Super, C.M. 1981: Social Roles and Moral
Reasoning: A Case Study in a Rural African Community. Developmental Psychology, 17, 595-603.
0 Blachell Publishers Ltd. 1996
Deontic Reasoning Modules
187
Harris, P.L. and Nufiez, M. in press: Understanding of Permission Rules by Preschool Children. Child Development.
Henle, M. 1962: On the Relation Between Logic and Thinking. Psychological
Review, 69, 366-78.
Hilpinen, R. 1971: Deontic Logic: lntroducto y and Systematic Readings. Boston:
Reidel / Kluwer.
Hilpinen, R. 1981: New Studies in Deontic Logic: Norms, Actions, and the Foundations
of Ethics. Boston: Reidel/Kluwer.
Humphrey, N.L.K. 1976: The Social Function of Intellect. In P.P.G. Bateson and
R.A. Hinde (eds), Growing Points in Ethology. Cambridge University Press.
Johnson-Laird, P.N., Legrenzi, P. and Legrenzi, M.S. 1972: Reasoning and a Sense
of Reality. British journal of Psychology, 63, 395400.
Johnson-Laird, P.N. and Byrne, R.M.J. 1991: Deduction. Hillsdale, N.J.: Erlbaum.
Jolly, A. 1966: Lemur Social Behavior and Primate Intelligence. Science, 153,
501-06.
Keil, F.C. 1994: Explanation, Association, and the Acquisition of Word Meaning.
Lingua, 92, 169-96.
Komatsu, L.K. and Galotti, K.M. 1986: Children's Reasoning About Social, Physical, and Logical Regularities: A Look at Two Worlds. Child Development, 57:
413-20.
Krebs, J.R. and Dawkins, R. 1984: Animal Signals: Mind Reading and Manipulation. In J.R. Krebs and N.B. Davies (eds), Behavior Ecology: A n Integrated
Approach. Oxford: Basil Blackwell.
Kummer, H. 1988: Tripartite Relations in Hamadryas Baboons. In R.W. Byrne
and A. Whiten (eds), Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford University Press.
Leslie, A.M. 1984: Spatiotemporal Continuity and the Perception of Causality in
Infants. Perception, 13: 287-305.
Leslie, A.M. 1987: Pretense and representation: The origins of 'theory of mind.
Psychological Review, 94, 412-26.
Leslie, A.M. and Frith, U. 1990: Prospect for a Cognitive Neuropsychology of
Autism: Hobson's Choice. Psychological Review, 97: 122-31.
Leslie, A.M. and Keeble, S. 1987 Do Six-month-old Infants Perceive Causality?
Cognition, 25, 265-88.
Light, P., Blaye, A., Gilly, M. and Girotto, V. 1989: Pragmatic Schemas and Logical Reasoning in 6- to 8-year-old Children. Cognitive Development, 4, 49-64.
Light, P., Girotto, V. and Legrenzi, P. 1990: Children's Reasoning on Conditional
Promises and Permissions. Cognitive Development, 5: 369-83.
Manktelow, K.I. and Over, D.E. 1990: Deontic Thought and the Selection Task.
In K.J. Gilhooly, M.T. Keane, R.H. Logie and G. Erdos (eds), Lines of Thinking
Volume 1. Chichester: Wiley.
Manktelow, K.I. and Over, D.E. 1991: Social Roles and Utilities in Reasoning
With Deontic Conditionals. Cognition, 39: 85-105.
Manktelow, K.I. and Over, D.E. 1995: Deontic Reasoning. In S.E. Newstead and
J. St. €5. Evans (eds), Perspectives on Thinking and Reasoning. Englewood Cliffs,
N.J.: Erlbaum.
Maynard Smith, J. 1982: Evolution and the Theoy of Games. Cambridge University Press.
McGonigle, B.O. and Chalmers, M. 1977: Are Monkeys Logical? Nature 267,
694-696.
0 Blackwell Publishers Ltd. 1996
188
Mind 6 Language
Much, N.C. and Schweder, R.A. 1978 Speaking of Rules: The Analysis of Culture
in Breach. New Directions for Ch Id Development, 2, 19-39.
Oaksford, M. and Chater, N. 1993: Reasoning Theories and Bounded Rationality.
In K.I. Manktelow and D.E. Over (eds), Rationality. London: Routledge.
Oaksford, M. and Chater, N. 1994: A Rational Analysis of the Selection Task as
Optimal Data Selection. Psychological Review, 101, 608-31.
Osherson, D.N. 1974: Logical Abilities in Children, Volumes 1 and 2. Hillsdale,
N.J.: Erlbaum.
Osherson, D.N. 1975: Logical Abilities in Children, Volumes 3 and 4. Hillsdale,
N.J.: Erlbaum.
Over, D.E. and Manktelow, K.I. 1993: Rationality, Utility, and Deontic Reasoning.
In K.I. Manktelow and D.E. Over (eds), Rationality. London: Routledge.
Packer, C. 1977: Reciprocal Altruism in Olive Baboons. Nature, 265, 441-3.
Platt, R.D. and Griggs, R.A. 1993 Darwinian Algorithms and the Wason Selection
Task: A Factorial Analysis of Social Contract Selection Task Problems. Cognition, 48, 163-92.
Politzer, G. and Nguyen-Xuan, A. 1992: Reasoning About Conditional Promises
and Warnings: Darwinian Algorithms, Mental Models, Relevance Judgments, or pragmatic schemas? Quarterly Journal of Experimental Psychology,
44A, 401-21.
Pope Edwards, C. 1982: Moral Development in Comparative Cultural Perspective. In D.A. Wagner and H.W. Stevenson (eds), Cultural Perspectives on Child
Development. San Francisco: W.H. Freeman.
Popper, K.R. 1959: The Logic of Scientific Discovery. London: Hutchinson.
PrudHomme, J. and Chapais, B. 1993: Aggressive Interventions and Matrilineal
Dominance Relations in Semifree-ranging Barbary Macaques. Primates, 34,
271-83.
Putnam, H. 1974: The ’Corroboration’ of Theories. In P.A. Schilpp (ed), The Philosophy of Karl Popper, Volume 1. LaSalle, IL.: Open Court.
Quine, W.V.Q. 1953: Two Dogmas of Empiricism. In W.V.Q. m i n e (ed.), From
a logical point of view. Cambridge, M.: Harvard University Press.
Reich, S.S. and Ruth, P. 1982: Wason’s Selection Task Verification, Falsification,
and Matching. Brifish Journal of Psychology, 73, 395-405.
Reilly, J.S. 1986: The Acquisition of Temporals and Conditionals. In E. Traugott,
A. ter Meulen, J.S. Reilly and C.A. Ferguson (eds), On Conditionals. Cambridge University Press.
Revlin, R. and Leirer, V.O. 1978: The Effects of Personal Biases on Syllogistic
Reasoning: Rational Decision from Personalized Representations. In R.
Revlin and R.E. Mayer (eds), Human Reasoning. Washington DC: Winston & Sons.
Rips, L.J. 1983: Cognitive Processes in Propositional Reasoning. Psychological
Review, 90, 38-71.
Rips, L.J. 1994 The Psychology of Proof. Cambridge, MA.: MIT Press.
Riss, D.C. and Goodall, J. 1977 The Recent Rise to the Alpha-rank in a Population
of Free-living Chimpanzees. Folia Primatologica, 27, 134-51.
Roberge, J.J. 1982: Linguistic Factors in Conditional Reasoning. Quarterly Journal
of Experimental Psychology, 34A, 275-84.
Rumain, B., Connell, J. and Braine, M.D.S. 1983: Conversational Comprehension
Processes are Responsible for Reasoning Fallacies in Children as Well as
Adults: IF is Not the Bioconditional. Developmental Psychology, 19, 471-81.
0 Blackwell Publishers Ltd. 2996
Deontic Reasoning Modules
189
Scribner, S. 1975 Recall of Classical Syllogisms: A Cross-cultural Investigation
of Error on Logical Problems. In R.J. Falmagne (ed.), Reasoning: Representation and Process in Children and Adults. Hillsdale, N.J.: Erlbaum.
Seyfarth, R.M. 1976 Social Relationships Among Adult Female Monkeys. Animal
Behavior, 24, 917-38.
Seyfarth, R.M. 1981: Do Monkeys Rank Each Other? Behavioral and Brain Science,
4,447-8.
Seyfarth, R.M. and Cheney, D.L. 1984: Grooming, Alliances, and Reciprocal
Altruism in Vervet Monkeys. Nature, 308, 541-3.
Shannon, C.E. and Weaver, W. 1949: The Mathematical Theory of Communication.
Urbana: University of Illinois Press.
Shultz, T.R. 1982: Causal Reasoning in the Social and Nonsocial Realms. Canadian
Journal of Behavioral Science, 14, 307-22.
Seligman, M.E.P. 1970: On the Generality of the Law of Learning. Psychological
Reviezu, 77, 406-18.
Smith, P.K. 1988: The Cognitive Demands of Children's Social Interactions With
Peers. In R.W. Byrne and A. Whiten (eds), Machiavellian Intelligence: Social
Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford
University Press.
Smuts, B. 1985: Sex and Friendship in Baboons. Chicago: Aldine Publishing Co.
Snarey, J.R. 1985: Cross-cultural Universality of Social-Moral Development: A
Critical Review of Kohlbergian Research. Psychological Bulletin, 97, 202-32.
Spelke, E. 1994: Initial Knowledge: Six Suggestions. Cognition, 50, 431-45.
Thompson, V.A. 1994; Interpretational Factors in Conditional Reasoning. Memory
and Cognition, 22, 742-58.
Thompson, V.A. 1995: Conditional Reasoning: The Necessary and Sufficient Conditions. Canadian journal of Experimental Psychology, 49, 1-60.
Thompson, V.A. and Mann, J.M. 1995: Perceived Necessity and Sufficiency
Explains the Dissociation Between Logic and Meaning: The Case of 'Only
If'. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21,
1-14.
Tooby, J. and Cosmides, L. 1989 The Logic of Threat: Evidence for Another Cognitive Adaptation? Paper Presented at the Human Behavior and Evolution
Society, Evanston, IL.
Trivers, R.L. 1971: The Evolution of Reciprocal Altruism. Quarterly Rmiew of
Biology, 46, 35-57.
Tronick, E.Z., Als, H., Adamson, L., Wise, S. and Brazelton, T.B. 1978: The
Infant's Response to Entrapment Between Contradictory Messages in Faceto-Face Interaction. journal of the American Academy of Child Psychiatry, 17,
1-13.
Uehara, S., Hiraiwa-Hasegawa, M., Hosaka, K. and Hamai, M. 1994: The Fate of
Defeated Alpha Male Chimpanzees in Relation to Their Social Networks.
Primates 35, 49-55.
Vandell, L. and Wilson, K.S. 1987 Infant's Interactions with Mother, Sibling, and
Peer: Contrasts and Relations Between Interaction Systems. Child Development, 58, 176-86.
Van Duyne, P.C. 1974 Realism and Linguistic Complexity in Reasoning. British
Journal of Psychology, 65, 59-67.
Wason, P.C. 1968: Reasoning about a Rule. Quarterly Journal of Experimental Psyc h o l o ~20,
, 273-81.
0 Blackwell Publishers Lfd. 1996
190
Mind 6 Language
Wason, P.C. and Johnson-Laird,P.N. 1972: Psychology of Reasoning: Structure and
Content. London: Batsford.
Wason, P.C. and Shapiro, D. 1971: Reasoning and Mental Representation. Quarterly Journal of Experimental Psychology, 36A, 597-610.
Whiten, A. and Byme, R.W. 1988a: Taking (Machiavellian) Intelligence Apart:
Editorial. In R.W. Byrne and A, Whiten (eds), Machiavellian Intelligence: Social
Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford
University Press.
Whiten, A. and Byrne, R.W. 1988b: The Manipulation of Attention in Primate
Tactical Deception. In R.W. Byme and A. Whiten (eds), Machiavellian intelligence: Social expertise and the evolution of intellect in monkeys, apes, and humans.
Oxford University Press.
Wiener, N. 1948: Cybernetics. New York: Wiley.
Wilkins, M.C. 1928: The Effect of Changed Material on the Ability to do Formal
Syllogistic Reasoning. Archives of Psychology, 16, no. 102.
Wimmer, H. and Perner, J. 1983: Beliefs about Beliefs: Representation and Constraining Function of Wrong Beliefs in Young Children’s Understanding of
Deception. Cognition, 13, 103-28.
Woodruff, G. and Premack, D. 1979: Intentional Communication in the Chimpanzee: The Development of Deception. Cognition, 7, 333-62.
0 Blackwell Publishers Ltd. 1996